Skip to content

FAQ

Models, setup, billing and the service itself. For setup steps, see the docs.

What is OpusVIP?

OpusVIP gives developers the Claude Opus lane that production projects already run on every day: the same resource pool and the same routing, unchanged. It is tuned for Claude Opus and built for Claude Code. For how a relay works, see What is a Claude Code relay?

Is OpusVIP an official Anthropic service?

No. OpusVIP is an independent third-party service. It is not affiliated with, authorized by or partnered with Anthropic. Claude and Opus are trademarks of Anthropic.

Are the models genuine? How can I check?

The model you request is the model that handles the request. The console logs every request with its model and token usage, so you can check at any time.

Which models are available?

Four: the flagship claude-opus-5-5, the most capable claude-fable-5-1, and two partners, claude-sonnet-5-5 and claude-haiku-5-5. All four have a 1M context window and up to 128K output tokens (Haiku 5.5 prompts over 100K tokens are billed at Anthropic's long-context rate). Full prices are on Models and pricing.

Which clients are supported? Can I use an SDK or curl?

OpusVIP is built for the official Claude Code CLI only. Requests from other clients, including SDKs and direct curl calls, are not accepted.

How does billing work?

Pay as you go in credits, with no monthly fee. 1 credit = $0.01 of usage at Anthropic's list prices, and 100 credits ($1 of usage) cost $0.15, so you pay 15% of list price. Every model is billed at 1x list price. Per 1M tokens: Opus 5.5 input 400 credits (you pay $0.60), output 2,000 ($3.00), cache read 20; Fable 5.1 input 1,000, output 5,000, cache read 25; Sonnet 5.5 input 200, output 1,000, cache read 10; Haiku 5.5 input 10, output 50, cache read 1 (prompts over 100K tokens use the long-context rate: input 50, output 250). Full table on Models and pricing.

How are credits calculated?

1 credit = $0.01 of usage at list price, and 100 credits cost $0.15. Each request is priced in USD at Anthropic's list prices, then multiplied by 100 to give the credits deducted. For example, a Claude Code task that would cost $0.90 at list price deducts 90 credits, which cost you $0.135. The console usage log shows the credits deducted for every request.

How do I switch models in Claude Code?

Once the two environment variables, ANTHROPIC_BASE_URL and ANTHROPIC_AUTH_TOKEN, are set, Claude Code starts on Opus 5.5. Type /model opus, /model fable, /model sonnet or /model haiku in a session to switch, or pass the --model flag at startup. See Choose and switch models.

Is prompt caching supported?

Yes. Cache writes and reads are billed at Anthropic's cache rates: a 5-minute write costs 1.25x the input price, a 1-hour write 2x, and a cache read far less than regular input. Claude Code caches by default, and the console log shows the cache read and write tokens for every request.

Are failed requests billed?

No. Upstream errors and requests that don't complete successfully are never charged.

How do I top up?

Top up in the console with Stripe: card, Alipay or WeChat Pay (Apple Pay and Google Pay work too). $0.15 buys 100 credits ($1 of usage at list price), and the minimum top-up is $1.50 (1,000 credits). Sign up for the console (Google sign-in supported) and top up there.

How can I check every charge?

The console usage log records the time, model, input / output / cache tokens and the amount charged for every request.